expertiseanalyses interventionsapproach aboutglossary FR let's talk
· strategic analysis · omer taki · march 2026

Many AI strategies fail
before they even
exist.

Most AI strategies fail not in execution, but in the quality of the decision framework that precedes it. An AI strategy is not a list of use cases. It is a decision framework on value, dependencies, control and responsibility.

· definition

A strategic AI decision is a decision whose consequences create dependencies or commit the organisation's trajectory for 5 to 10 years: infrastructure choices, LLM integration in critical processes, data policy, deployment of a high-risk EU AI Act system.

For French executive committees, the ability to correctly arbitrate what is at stake before deciding is the differentiating factor.

AI changes who decides, how, on what data, with what constraints and in what power relationship. Most organisations believe they have an AI strategy. They have a collection of initiatives, POC, tools and subscriptions. The problem is not execution. It is the quality of decisions made upstream.

An AI strategy is not a list of use cases

A list of use cases answers: what can we do with AI? An AI strategy answers different questions: which capabilities must we own, which dependencies are acceptable, which risks do we escalate to the board?

AI changes the structure of decision

Before AI, decisions were made by humans who could account for them. With AI, some decisions are pre-oriented, accelerated or automated. If you cannot explain, contest, or reproduce the decision: you are ceding power.

Three structural failures

1. Decisions made at the wrong level

Structuring AI decisions are made by business or IT teams, without board visibility, without escalation criteria, without doctrine.

2. Incomplete criteria

AI decisions are evaluated on immediate ROI or technical feasibility, without measuring dependencies created, control ceded or regulatory obligations generated.

3. Absence of decision architecture

No framework exists to distinguish what belongs to operations and what belongs to strategy. A poorly calibrated decision does not create a small inefficiency. It creates an irreversibility.

The wrong question. And the right one.

Wrong question: "how do we deploy AI faster?"

Right question: "which decisions are we making today that will structure what we can do in 5 years? Not more tools. Not more pilots. A complete reading of what is at stake."

Three questions to assess your exposure:
who decides on AI in your organisation?
with what criteria? at what level?
· tointelligence

Three questions to assess your exposure.
Answer them before your next board meeting.

We structure the AI decision framework at executive level. Exclusively executive committees.

let's talk